COVID-19: Estimates suggest COVID-19 transmission rates are highly seasonal

COVID-19: Estimates suggest COVID-19 transmission rates are highly seasonal. Spencer EA, Heneghan C.

Published on June 27, 2020

Reference Carleton T, Meng KC. Causal empirical estimates suggest COVID-19 transmission rates are highly seasonal medRxiv 2020
Study type
Country Global
Funding Details Non Reported
Transmission mode Meteorological
Exposures Temperature, Humidity, Population density

Bottom Line

Seasonal temperature is associated with COVID-19 transmission globally, with 1°C increase in local temperature associated with 13% fewer cases.

Evidence Summary

Adjusting for precipitation, humidity, population density and public health intervention, the association between weather temperature and COVID-19 cases was estimated as:

each 1C increase in local temperature was associated with 13%, fewer COVID-19 cases per 1 million people by (95% CI -4% to -21%).

This result was robust to controlling for precipitation and specific humidity, neither of which exhibited statistically significant effects (although point estimates on humidity were negative, consistent with prior evidence for influenza).

What did they do?

The study estimated the relationship between local temperature and COVID-19 transmission, using a global sample of 166,686 confirmed new COVID-19 cases from 134 countries from 22nd January 2020 to 15th March 2020, aiming to control for local public health interventions, surrogate for UV exposure, and population densities.

Data on the following were collected: daily temperature, precipitation, specific humidity for every 0.25◦ latitude by 0.25◦ longitude pixel of the planet, generated by a climate reanalysis model in near real-time.

Using population weights pixel-level weather variables were aggregated to country-level reports of COVID-19 cases.  A global function between temperature and new cases of COVID-19 per 1 million people was estimated.

The study used the ERA5 reanalysis product from the European Centre for Medium-Range Weather Forecasts:!/home: The ERA5  is freely available and functions as a one-stop-shop to explore climate data. 

Study reliability

This is a large sample and the methods aim to avoid major confounding by environmental and socioeconomic influences.

Only confirmed cases were analysed which may under-estimate the magnitude of the link between infection and local climatic conditions. Countries around the world also have different testing capacity, making under-reporting heterogeneous and estimates prone to bias.

Clearly defined setting Demographic characteristics described Follow-up length was sufficient Transmission outcomes assessed Main biases are taken into consideration
Yes No Unclear Unclear Yes

What else should I consider?

Temperature is correlated with many often unobservable potential confounding factors. Cross-sectional comparisons may not have a causal interpretation. e.g., countries that are cooler on average also tend to have higher income per capita which may affect the number of new COVID-19 cases by enabling more testing and hospitalizations.

About the authors

Carl Heneghan

Carl Heneghan

Carl is Professor of EBM & Director of CEBM at the University of Oxford. He is also a GP and tweets @carlheneghan. He has an active interest in discovering the truth behind health research findings

Elizabeth Spencer

Elizabeth Spencer

Dr Elizabeth Spencer; MMedSci, PhD. Epidemiologist, Nuffield Department for Primary Care Health Sciences, University of Oxford.

Tom Jefferson

Tom Jefferson

Tom Jefferson is a senior associate tutor and honorary research fellow, Centre for Evidence-Based Medicine, University of Oxford.